LiDAR: measuring shape and distance directly from the round trip of light

LiDAR times how long a laser pulse takes to hit a target and return: distance = speed of light × round-trip time ÷ 2. Light travels about 30 cm per nanosecond, so one nanosecond of round-trip time is about 15 cm of range. Very fine timing electronics are the heart of the instrument.

The last row matters most. A drone LiDAR measures only range and direction from the aircraft; map coordinates depend on the accuracy of the aircraft’s position and attitude. Where GNSS does not reach, that trajectory must instead be estimated from the surroundings with SLAM.

LiDAR produces a point cloud: each point carries coordinates (X, Y, Z), intensity, return number and time. ASPRS LAS (LAZ when compressed) is the common standard format. Add colour from a camera and you have a coloured point cloud.

• Point density: points per square metre. Thin wires and small members need density and good incidence.
• Classification: sort points into ground, buildings, vegetation, conductors, towers. Automatic classification followed by human editing is normal.
• DTM & DSM: bare-earth height (DTM) versus surface height including trees and buildings (DSM). Their difference is canopy height.
• Cross-sections: for tunnels and pipes, slice the cloud to check deformation of the inner profile.

One of LiDAR’s most mature inspection uses is power lines. It measures clearance between conductors and trees, buildings or ground along long corridors in one pass. Multiple returns reach the ground through foliage, so it is much faster than foot or helicopter patrols in forests.

The same approach works for railway catenary and lineside trees. Flying near conductors can disturb the compass and links, so keep a generous distance. Strong electromagnetic fields

Inside tunnels, mines, storage tanks and boilers there is no GNSS. LiDAR scans are matched against each other to estimate motion and build the map at the same time (SLAM). Working in total darkness is an advantage cameras lack. Caged, collision-tolerant drones carry LiDAR for this (e.g. Flyability Elios 3).

For the environments in detail, see GNSS-denied, confined spaces, darkness and dust & smoke.

The “range accuracy” in a LiDAR datasheet is for the sensor alone. Map accuracy stacks trajectory error, boresight (mounting angle) error and beam divergence on top. Even good drone LiDAR results are typically in the range of a few centimetres — and should always be verified against check points.

In practice carrying both is increasingly common: LiDAR builds the geometric skeleton, camera images supply colour and detail. Crack widths come from photos; deformation and clearance from LiDAR.

How to keep a good-looking 3D model separate from survey-grade accuracy is walked through in separating appearance from accuracy in drone survey (Airffic Flight Institute).

• Fine surface defects: thin cracks and rust colour do not show in a point cloud.
• Underwater: common near-infrared LiDAR is absorbed by water; bathymetry needs green-laser systems, which struggle in turbid water.
• Dust, smoke, rain: particles cause false points and gaps.
• SLAM in uniform spaces: in long tunnels, keeping distances right without targets is hard.
• Decisions: clearance criteria and cutting priorities are set by the asset owner.

By 2035 LiDAR will be lighter and cheaper and likely standard alongside the camera on most inspection aircraft. Power-line corridors are captured on the same route every year, and cutting schedules are proposed from predicted tree growth and conductor sag. Inside shafts and tanks, SLAM clouds are overlaid on the last survey and only wall deformation is coloured. This is the editors’ view.

Even then, accuracy will be guaranteed by old-fashioned check points and recorded conditions. The more we can measure, the more weight falls on showing how far the data can be trusted.